Two Cellular Mechanisms for Gamma-Globin Gene and Protein Silencing in Humans.

Although the genetic processes responsible for gamma-globin gene and protein silencing are not known, the prevailing model is that gamma-globin silencing results from a gradual change within a single hematopoietic cell lineage that is governed by intrinsic properties of the cells. In order to provide a more complete characterization of the silencing phenomenon, we studied globin expression patterns directly from clinical samples using single-cell, quantitative PCR, and globin protein phenotyping. We collected blood samples from untransfused donors: umbilical cords (n=3), infants (n=11; ages 1 day to 35 months), and adults (n=3). All samples were maintained at 4°C and analyzed within 72 hours. Flow cytometry (30,000 cells per donor) and HPLC analyses were used for globin protein phenotyping. For globin gene expression, we identified reticulocytes using a strategy that required no membrane permeabilization, and sorted them as single cells directly into lysis buffer. Oligo-dT reverse transcription of mRNA was followed by real-time PCR quantitation. Globin cDNA copy numbers were calculated using standard curves from serial dilutions of a plasmid DNA. We analyzed approximately 1000 single-cell quantitative PCR amplifications for gamma- and beta-globin gene expression. In cord blood, we detected both gamma- and beta-globin gene expression in 97.4% (112/115) of the reticulocytes. The average gamma-globin cDNA copy number was 1870±1390 copies, compared with an average beta-globin cDNA copy number of 2181±2138 copies per reticulocyte. HbF and HbA were also detected in >95% of the cord blood erythrocytes. In the adult samples, HbF was detected in <5% of the circulating erythrocytes and gamma-globin gene expression in only 1.5% (3/206) of the reticulocytes. The average gamma-globin cDNA copy number in the minor population of gamma(+) adult reticulocytes was 468±198 copies, and the average beta-globin cDNA copy number in the beta(+) adult reticulocytes was 3869±3733 copies. Compared with the relatively monotonous patterns of gamma-globin gene and protein expression in cord and adult blood, we clearly detected an age-based fluctuation between those patterns in the infant blood samples. During the first three years of life, a gradual loss in the level of gamma-globin gene and protein expression was identified among the gamma(+)beta(+) reticulocytes and the HbF(+)HbA(+) erythrocytes. In addition, discrete populations of gamma(−)beta(+)reticulocytes and HbF(−)HbA(+) erythrocytes were detected. Rapid expansion of those gamma-silenced populations became apparent soon after birth. Within four months, the proportion of gamma-silenced cells eclipsed that of the gamma(+)beta(+) cells to become the predominant population. By three years after birth, the two cell populations essentially merged to become a single, gamma-silenced population similar to that found in adults. These data suggest two cellular mechanisms for gamma-globin silencing in humans: 1) a gradual loss in gamma-globin expression in the gamma(+)beta(+) cells beginning prior to delivery and continuing during infancy, and 2) replacement of the gamma(+)beta(+) cells with a population of gamma-silenced cells that rapidly accumulate after birth, possibly in response to the dramatic increase in oxygenation or other environmental changes.